Question

batchelor correctExperiments() returns Error in eval(subscript, envir = eframe, enclos = eframe) : '...' used in an incorrect context

0

Entering edit mode

pcantalupo ▴ 10

@pcantalupo-8617

Last seen 5 months ago

United States

I'm trying to follow the batchelor vignette using my own data. I have two batches "A" (2 samples) and "P" (3 samples). I combined the A samples using Reduce and separately used Reduce for the 3 P samples. Then I performed normalization and feature selection separately on each SCE object. When running correctExperiments with no correction: correctExperiments(sceA, sceP, PARAM=NoCorrectParam()) and subsequent PCA/TSNE, it works. But when running it with the default PARAM (fastMNN) correctExperiments(sceA, sceP, subset.row=chosen.hvgs) I get the following error after about 5 minutes:

Error in eval(subscript, envir = eframe, enclos = eframe) : 
  '...' used in an incorrect context

The code that I ran is below:

universe <- Reduce(intersect, lapply(all.sce, rownames))

abmmc_samples = lapply(all.sce[1:2], "[", i=universe)
abmmc_combined = Reduce(combineCols, lapply(abmmc_samples, function(a) {return(a)}))

pbmmc_samples = lapply(all.sce[3:5], "[", i=universe)
pbmmc_combined = Reduce(combineCols, lapply(pbmmc_samples, function(p) {return(p)}))

out = multiBatchNorm(abmmc_combined, pbmmc_combined)
sceA = out[[1]]
sceP = out[[2]]

decA <- modelGeneVar(sceA)
decP <- modelGeneVar(sceP)
combined.dec <- combineVar(decA, decP)
chosen.hvgs <- getTopHVGs(combined.dec, n=1000)

# runs OK with no correction
combined <- correctExperiments(sceA, sceP, PARAM=NoCorrectParam())  # no correction
combined = fixedPCA(combined, subset.row=chosen.hvgs, rank = 10) 
combined <- runTSNE(combined, dimred="PCA")
plotTSNE(combined, colour_by="group")

# error with correction
mnn <- correctExperiments(sceA, sceP, subset.row=chosen.hvgs)  # fastMNN correction
#Error in eval(subscript, envir = eframe, enclos = eframe) : 
# '...' used in an incorrect context

What am I doing wrong? Thank you

batchelor • 1.4k views

ADD COMMENT • link 22 months ago pcantalupo ▴ 10

0

Entering edit mode

sorry, forgot to post sessionInfo()

> sessionInfo()
R version 4.1.2 (2021-11-01)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows 10 x64 (build 19043)

Matrix products: default

locale:
[1] LC_COLLATE=English_United States.1252  LC_CTYPE=English_United States.1252   
[3] LC_MONETARY=English_United States.1252 LC_NUMERIC=C                          
[5] LC_TIME=English_United States.1252    

attached base packages:
[1] stats4    stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
 [1] scater_1.22.0               ggplot2_3.3.5               BiocParallel_1.28.3        
 [4] scran_1.22.1                scuttle_1.4.0               batchelor_1.10.0           
 [7] SingleCellExperiment_1.16.0 SummarizedExperiment_1.24.0 Biobase_2.54.0             
[10] GenomicRanges_1.46.1        GenomeInfoDb_1.30.1         IRanges_2.28.0             
[13] S4Vectors_0.32.4            BiocGenerics_0.40.0         MatrixGenerics_1.6.0       
[16] matrixStats_0.62.0         

loaded via a namespace (and not attached):
 [1] viridis_0.6.2             edgeR_3.36.0              BiocSingular_1.10.0      
 [4] viridisLite_0.4.0         DelayedMatrixStats_1.16.0 assertthat_0.2.1         
 [7] statmod_1.4.36            BiocManager_1.30.17       dqrng_0.3.0              
[10] GenomeInfoDbData_1.2.7    vipor_0.4.5               ggrepel_0.9.1            
[13] pillar_1.7.0              lattice_0.20-45           glue_1.6.2               
[16] limma_3.50.3              beachmat_2.10.0           XVector_0.34.0           
[19] colorspace_2.0-3          Matrix_1.4-1              pkgconfig_2.0.3          
[22] zlibbioc_1.40.0           purrr_0.3.4               scales_1.2.0             
[25] ScaledMatrix_1.2.0        Rtsne_0.16                ResidualMatrix_1.4.0     
[28] tibble_3.1.6              generics_0.1.2            ellipsis_0.3.2           
[31] withr_2.5.0               cli_3.2.0                 magrittr_2.0.3           
[34] crayon_1.5.1              fansi_1.0.3               bluster_1.4.0            
[37] beeswarm_0.4.0            tools_4.1.2               lifecycle_1.0.1          
[40] munsell_0.5.0             locfit_1.5-9.5            cluster_2.1.3            
[43] DelayedArray_0.20.0       irlba_2.3.5               compiler_4.1.2           
[46] rsvd_1.0.5                rlang_1.0.2               grid_4.1.2               
[49] RCurl_1.98-1.6            BiocNeighbors_1.12.0      igraph_1.3.1             
[52] bitops_1.0-7              gtable_0.3.0              DBI_1.1.2                
[55] R6_2.5.1                  gridExtra_2.3             dplyr_1.0.8              
[58] utf8_1.2.2                metapod_1.2.0             ggbeeswarm_0.6.0         
[61] parallel_4.1.2            Rcpp_1.0.8.3              vctrs_0.4.1              
[64] tidyselect_1.1.2          sparseMatrixStats_1.6.0

ADD REPLY • link 22 months ago pcantalupo ▴ 10

0

Entering edit mode

According to this github issue (https://github.com/LTLA/batchelor/issues/32), you may need to update both batchelor and DelayedArray packages

ADD REPLY • link 22 months ago Basti ▴ 780

0

Entering edit mode

Thank you Basti for the tip. However after upgrading batchelor and DelayedArray

> pacman::p_version('batchelor')
[1] ‘1.13.1’
> pacman::p_version('DelayedArray')
[1] ‘0.23.0’

I get a new error when running mnn <- correctExperiments(sceA, sceP, subset.row=chosen.hvgs): Error: cannot subset by character when rownames are NULL. The rownames are not NULL:

> str(rownames(sceA))
 chr [1:11092] "ENSG00000225880" "ENSG00000188976" "ENSG00000188290" "ENSG00000187608" "ENSG00000131591" ...
> str(rownames(sceP))
 chr [1:11092] "ENSG00000225880" "ENSG00000188976" "ENSG00000188290" "ENSG00000187608" "ENSG00000131591" ...

However, good news...I found a way around this. I combined my SCE objects into one object sceAP = combineCols(sceA, sceP) and when running mnn <- correctExperiments(sceAP, batch = sceAP$group, subset.row=chosen.hvgs), it works!

Hopefully somebody has more explaination of what is going on but I think all is OK using this hack.

ADD REPLY • link 22 months ago pcantalupo ▴ 10

score 1 · Answer 1 · 2022-06-18

1

Entering edit mode

Aaron Lun ★ 28k

@alun

Last seen 31 minutes ago

The city by the bay

As @Basti said, the ... used in an incorrect context should be fixed by the Github issue. So if you have batchelor 1.12.1 or higher, you should be fine.

I don't know about the rownames are NULL error though. You'll have to post traceback immediately after encountering the error, e.g., with traceback(max.lines=10).

ADD COMMENT • link 22 months ago Aaron Lun ★ 28k

0

Entering edit mode

Here is the output:

> mnn <- correctExperiments(sceA, sceP, subset.row=chosen.hvgs)
Error: cannot subset by character when rownames are NULL
In addition: Warning messages:
1: In .accumulate_rowdata(all.rrw) :
  ignoring non-identical 'Symbol' field in 'rowData'
2: In .accumulate_rowdata(all.rrw) :
  ignoring non-identical 'mean' field in 'rowData'
3: In .accumulate_rowdata(all.rrw) :
  ignoring non-identical 'detected' field in 'rowData'

> traceback(max.lines=10)
11: stop(wmsg(...), call. = FALSE)
10: .subscript_error("cannot subset by character when ", what, " are NULL")
9: NSBS(i, x, exact = exact, strict.upper.bound = !allow.append, 
       allow.NAs = allow.NAs)
8: NSBS(i, x, exact = exact, strict.upper.bound = !allow.append, 
       allow.NAs = allow.NAs)
7: normalizeSingleBracketSubscript(i, x, allow.NAs = TRUE, as.NSBS = TRUE)
6: extractROWS(x, i)
5: extractROWS(x, i)
4: combined.rd[subset.row, , drop = FALSE]
3: combined.rd[subset.row, , drop = FALSE]
2: .create_fresh_combined_sce(x, merged, subset.row = subset.row, 
       correct.all = correct.all, combine.assays = combine.assays, 
       combine.coldata = combine.coldata, include.rowdata = include.rowdata)
1: correctExperiments(sceA, sceP, subset.row = chosen.hvgs)

Thank you for your help.

ADD REPLY • link 22 months ago pcantalupo ▴ 10

1

Entering edit mode

Should be patched in batchelor 1.12.2; give it a few days to make its way through BioC's build system, or you can install directly from the RELEASE_3_15 branch on the GitHub repository.